“Schedule”版本间的差异

来自cslt Wiki
跳转至： 导航、 搜索
（→‎Month Summary）
（→‎Month Summary）
第165行： 第165行：
 
|Aodong Li ||  
 
|Aodong Li ||  
 
*Code:  
 
*Code:  
−
1. sequence-to-sequence model with attention mechanism for chatting model and lyrics generation in theano and tensorflow
+
  1. sequence-to-sequence model with attention mechanism for chatting model and lyrics generation in theano and tensorflow
−
2. Rare word embedding on word2vec and the related in C
+
  2. Rare word embedding on word2vec and the related in C
−
3. Help Lantian for ICASSP
+
  3. Help Lantian for ICASSP
 
*Paper sharing: two shared papers, one for rare word embedding, one for bilingual word embedding
 
*Paper sharing: two shared papers, one for rare word embedding, one for bilingual word embedding
 
*ML book: first version of chapter 2, linear model
 
*ML book: first version of chapter 2, linear model

2016年8月30日 (二) 08:59的版本

NLP Schedule

Members

Current Members

  • Yang Feng (冯洋)
  • Jiyuan Zhang （张记袁）
  • Aodong Li (李傲冬)
  • Andi Zhang (张安迪)
  • Ziwei Bai （白子薇）
  • Aiting Liu (刘艾婷)
  • Shiyao Li （李诗瑶）

Former Members

  • Chao Xing (邢超)  : FreeNeb
  • Rong Liu (刘荣)  : 优酷
  • Xiaoxi Wang (王晓曦) : 图灵机器人
  • Xi Ma (马习)  : 清华大学研究生
  • Tianyi Luo (骆天一) ： phd candidate in University of California Santa Cruz
  • Qixin Wang (王琪鑫)  : MA candidate in University of California
  • DongXu Zhang (张东旭): --
  • Yiqiao Pan (潘一桥) ： MA candidate in University of Sydney


Work Progress

Daily Report

Date Person start leave hours status
2016/08/16 Ziwei Bai 9:30 17:30 7 learn kernel construction,write the first construction method.
Andy Zhang 9:30 17:30 7 prepare for the ML book, watch Wang's video of neural network
Aiting Liu 9:30 17:30 7 learn Wang's instructional video of Neural Network
Shiyao Li
2016/08/17 Ziwei Bai 9:00 18:00 8 rewrite 'the property of Gram matrix' and write the second kernel construction method
Andy Zhang 9:00 18:00 8 read materials about Radio basis function networks as well as self-organizing map
Aiting Liu 9:00 18:00 8 view the information related to the Hopfield network and Boltzmann machine
Shiyao Li
2016/08/18 Ziwei Bai 9:30 18:30 8 learn the 'propert of kernel function'&learn chaos' chatting model predict program
Andy Zhang 9:30 18:30 8 continue preparing for the MLbook
Aiting Liu 9:30 18:30 8 prepare for the presentation, learn RBF
Shiyao Li
2016/08/19 Ziwei Bai 9:00 18:00 8 finish the first edition of my work of ML book
Andy Zhang 9:00 18:00 8 learn Hopfield network, prepare for the writing
Aiting Liu 9:00 18:00 8 learn Andrew Ng's online course
Shiyao Li
2016/08/20 Ziwei Bai 9:40 18:50 8 perfect my work of the book
Andy Zhang
Aiting Liu
Shiyao Li
2016/08/22 Ziwei Bai 9:15 18:15 8 test Chatting model ,retrain word vector
Andy Zhang 9:15 18:15 8 start writing chapter 3, version 0.0.
Aiting Liu
Shiyao Li
2016/08/23 Ziwei Bai 9:00 18:15 8 1、retrain NRM model & word vec 2.review paper‘NRM for STC’
Andy Zhang 9:00 18:15 8 continue writing the ML book. mainly worked on MLP today
Aiting Liu
Shiyao Li 9:00 18:15 8 read the book Deep Learning and prepare the presentation of Probability Theory
2016/08/24 Ziwei Bai 9:05 18:05 8 test Chatting model 0-8,modify ASR Python Server Functional Specification
Andy Zhang 9:05 18:05 8 mainly wrote the section of MLP & RBF
Aiting Liu
Shiyao Li 9:05 18:05 8 finish the preparation of the presentation about Linear Algebra and Probability Theory
2016/08/25 Ziwei Bai 9:20 21:00 7.5 1、test Chatting Model 9-14，and upload to cvss。2、learn Bengio‘s dialogue system by hierarchical NN（except experiment and result）。
Andy Zhang 9:20 21:00 7.5 continue writing chapter3
Aiting Liu
Shiyao Li 9:20 21:00 7.5 revise the PPT and start to learn tensorflow
2016/08/26 Ziwei Bai 9:30 18:30 8 prepare for seminar,select 100 posts Suitable for dialogue from weibo data
Andy Zhang 9:30 18:30 8 continue writing chapter3
Aiting Liu
Shiyao Li 9:30 18:30 8 self-study tensorflow
2016/08/27 Ziwei Bai 9:30 20:30 10 1、filter the movie scipts data based on weibo data word vector (30W+pairs) and try to train Chatting Model with the dataset on the basis of model 31, but load model failed(maybe because the version of tensor flow not consistent).2、find the response Corresponding to the post selected yesterday and test the model (0-34) with it
Andy Zhang
Aiting Liu
Shiyao Li
2016/08/29 Ziwei Bai 9:20 18:00 7+ test the Chatting model,count the number of startswith <END> sentences,and natural sentences.
Andy Zhang 9:20 17:40 7+ continued working on chapter 3, did some revise work & tried to find ways to draw figures
Aiting Liu 9:20 18:20 8 prepare diagrams in the book
Shiyao Li 9:20 18:20 8 learn the CNN to solve MNIST using tensorflow
Aodong Li 10:30 16:30 5 Read shared paper at the week before the last week
2016/08/30 Ziwei Bai 9:15
Andy Zhang
Aiting Liu
Shiyao Li 9:15

Month Summary

People Summary
Yang Feng
  • surveyed the line works of neural parsing, neural Turing machine and variants of RBMs and fixed the research topic to jointly parsing and understanding with Turing machine
  • learned tensorflow
  • produced two versions of results of query paraphrase
Jiyuan Zhang
Aodong Li
  • Code: 
 1. sequence-to-sequence model with attention mechanism for chatting model and lyrics generation in theano and tensorflow
 2. Rare word embedding on word2vec and the related in C
 3. Help Lantian for ICASSP
  • Paper sharing: two shared papers, one for rare word embedding, one for bilingual word embedding
  • ML book: first version of chapter 2, linear model
  • Read several papers about RNN, language model, word2vec, sequence-to-sequence model
Ziwei Bai
  • finish part of chapter 5 kernel method
  • retrain & test Chatting Model
  • share 'the use of tensorflow'
Andy Zhang
  • shared the principle of word2vec
  • read several papers about comparison of sentence similarity
  • wrote chapter 3 neural network of the ML book
Aiting Liu
  • share two papers:

1.Intrinsic Subspace Evaluation of Word Embedding Representations [pdf] [slides]

2.A Sentence Interaction Network for Modeling Dependence between Sentences [pdf] [slides]

  • write chapter2 linear model
  • help Andy complete chapter3 neural network
Shiyao Li
  • prepare the data for question-pair task.
  • use SRILM to train the language model and Moses to train the task
  • share a paper about LSTM and do a presentation.

Time Off Table

Date Yang Feng Jiyuan Zhang Aodong Li Ziwei Bai Andy Zhang Aiitng Liu Shiyao Li
2016/8/15 8 hours
2016/8/16 8 hours
2016/8/17 8 hours
2016/8/18 2 hours 4 hours
2016/8/19 6 hours 8 hours
2016/8/22 8 hours 8 hours
2016/8/23 2 hours 8 hours
2016/8/24 8 hours
2016/8/25 8 hours
2016/8/26 5 hours 8 hours
2016/8/29 0.5 hours

Past progress

nlp-progress 2016/08

nlp-progress 2016/05-07

nlp-progress 2016/04

取自“http://cslt.org/mediawiki/index.php?title=Schedule&oldid=22189”